Voice dictation startup Willow has launched Atlas 1 , a new speech-to-text model it calls a "frontier" system for real-time transcription. The YC-backed company claims the model ou
Alibaba's Wan 2.7 Image Model Prioritizes Control Over Aesthetics April 06, 2026 Alibaba has released Wan 2.7 Image , a new generative model that prioritizes prompt adherence and structural control over raw aesthetic quality. While previous iterations like Wan 2.1 focused hea
AI video platform PixVerse has released V6, a major version update that pushes its generation model beyond single-clip novelty and into structured, multi-shot workflows. The update brin
A new open-source method called Seen2Scene can now complete partial 3D scans into full, coherent scenes. The system addresses a fundamental constraint in 3D digitization: photogrammetry, Li
SpatialLM is a new open-source language model that takes raw 3D point clouds as input and outputs structured, machine-readable descriptions of indoor spaces. Feed it a messy 3D scan, and the mod
PhotaLabs has launched Phota , a new AI photo generation model focused on high-fidelity personalized image creation. Available now through platforms like fal , the system is designed to s
Sync Labs has officially launched Sync-3, a new generation of its AI lipsync model built specifically to handle the edge cases that cause other tools to fail. While most current
FullCircle is a new open-source pipeline that produces clean 3D reconstructions from handheld 360-degree video shot on a consumer camera. Walk through a space with something like an Insta360 X4,
GaussianGPT is a new research tool that generates full 3D Gaussian splat scenes using the same next-token prediction approach that powers large language models. Feed it a partial room and it fill
AI tool developer PROMPTR Studios has released CRAFTR , a Chrome extension built to give filmmakers and visual artists a structured environment for writing prompts across AI generation platforms.
FreeOrbit4D , accepted at SIGGRAPH 2026, generates bullet-time and free-viewpoint camera effects from a single ordinary video — with no specialized capture setup and no train
ElevenLabs has launched Music Finetunes in ElevenCreative , a tool that lets users train a custom version of the ElevenLabs Music model on their own audio catalog. The result is a personalized AI
Maxon and Tencent Cloud announced a partnership to integrate Tencent's HY 3D Global AI engine directly into Cinema 4D , enabling artists to generate base 3D and UV models from text prompts or imag
Lightricks has released LTX-2.3 , a major upgrade to its open-source video generation model. The update focuses on core engine improvements rather than interface cha
Lightricks has released LTX Desktop , an open-source desktop application for generating videos with LTX models on local hardware. The app runs natively on Windows with NVIDIA GPUs (32GB VRAM
Luma's Uni-1 Unified Intelligence: One Model for Thinking and Rendering March 06, 2026 Luma Labs launched Uni-1 , their first unified understanding and generation model, combining understanding and generation in a single model for the first time. Rather than separating the ta
Adobe Firefly's video editor now includes Quick Cut , an AI feature that assembles raw footage into a structured first-draft edit. Describe what you want in natural language — "lengthy intervie
AI agents have been getting all the buzz — but Perplexity and Anthropic just said hold my beer. Both companies dropped updates that push squarely into agentic territory — persisten
Foundry shipped Nuke 17.0 with two major additions: native Gaussian Splat support and a USD-based 3D system moving out of beta. The release targets VFX artists working in c
Google Launches Lyria 3: Gemini's High-Fidelity AI Music Generator February 19, 2026 Google has released Lyria 3 , a high-fidelity AI music generator built directly into the Gemini app. According to Google's music generation overview , the tool transforms text or image pr
Moonshot AI launched Kimi Claw , a browser-native implementation of OpenClaw that removes the technical setup typically required to run autonomous AI agents. The service runs in a single browser tab with
Beeble has released SwitchX , the company's first video-to-video model, announced by CEO Hoon Kim. Unlike Beeble's existing SwitchLight tools, which focus on generating PBR maps for relighting, S
Seedance 2.0: ByteDance's AI Video Model Goes Multimodal with 15-Second Clips and Dual-Channel Audio February 13, 2026 ByteDance has launched Seedance 2.0, a significant upgrade to its AI video generation model that introduces a unified multimodal architecture capable of proc
Vizrt, known for its broadcast graphics technology, is integrating augmented reality capabilities directly into Zoom through two new products launching February 20, 2026 . The partners
